Thesis
News-sentiment operates on a clean, intuitive premise: buy constituents of its 24-stock large-cap universe when recent news sentiment turns positive, and exit when it turns negative. The universe spans mega-cap technology, financials, healthcare, consumer staples, and energy — companies that attract consistent press coverage, making them natural candidates for a sentiment-driven approach.
Recent Activity
Over the past week, the strategy has been running daily but finding no actionable signals. Scheduled runs on June 19–24 each executed zero trades, with the portfolio sitting at roughly $2,810 cash against a total value oscillating between $9,688 and $9,748. The most recent trade was a buy of 4 shares of UNH at $402.85 on June 18. Prior to that, BAC (35 shares at $55.93, June 12), MSFT (4 shares at $428.23, June 3), and AAPL (6 shares at $312.06, May 31) round out the recent trade log.
The signal drought is notable. Four consecutive daily runs without a trigger suggests either the sentiment data is presenting a neutral-to-mixed picture across the universe, or the threshold for signal confirmation is appropriately conservative — the distinction matters and is worth monitoring.
Validation & Performance
The headline backtest numbers are superficially encouraging: 0.19% total return, a 5 basis-point maximum drawdown, and a Sharpe ratio of 0.74 over 451 days. A Probabilistic Sharpe Ratio (PSR) of 0.923 means there is roughly a 92% probability that the strategy's Sharpe exceeds zero — a reasonable confidence threshold in isolation.
The cross-validation picture is considerably more sobering. Across four walk-forward folds spanning August 2024 to May 2026, only the final fold (December 2025 – May 2026) produced any trades at all. Folds 1 through 3 each recorded zero trades, zero return, and zero drawdown — not because the strategy was cautious, but because it apparently never triggered. All of the strategy's observable edge is compressed into a single 5-month window.
The Deflated Sharpe Ratio (DSR) of 0.551, which adjusts for multiple-testing across six trials, underscores the concern: less than 55% confidence that the measured edge is real after accounting for the number of parameter configurations tested. Validation status is formally failed.
Risks and Outlook
The central risk is low signal frequency masking an absence of signal. Two backtest trades and a handful of live paper trades is insufficient sample size to distinguish a legitimate edge from noise. The strategy also currently reports a win rate of 0, though this reflects the absence of closed positions rather than a string of losses.
The upside case is that the fund of record has barely deployed capital (roughly 71% remains in cash), so when — if — sentiment conditions align, the strategy has room to build a position. The fold-4 out-of-sample Sharpe of 1.5 is worth watching as live data accumulates.
For now, news-sentiment sits in a holding pattern: low risk, low activity, unproven edge. The next meaningful evaluation point is when it accumulates enough trades to generate statistically meaningful win-rate and Sharpe estimates independent of the backtest period.